StylAR
Restyle the visible scene. One face, six finishes, swapped on the live frame.
What it does
Hold the camera up. The feed restyles itself in place — the shape of your face holds, the texture changes. Cycle through the styles by tapping. The look follows your head as it moves, your hair as it falls, the corner of your jaw as you turn.
How it works
A face-tracking layer reads landmarks every frame. A mask layer projects the chosen style onto that landmark geometry. The result is composed on top of the live feed via DeepAR and Snap CameraKit. Recon — the experimental lab — runs the same pipeline on a still photograph for when you want to see how a style settles before you commit.
What it's built on
ARKit face tracking. DeepAR for the mask projection. Snap CameraKit for the catalog of styles. MaskKit for the occlusion math. Recon for the still-frame preview.